https://github.com/bekcsys/LLM-Chatbot
LLM Based ChatBot for Quantitative Carbon foorprint analysis
https://github.com/bekcsys/LLM-Chatbot
deepseek-r1 llama3 llms mistral
Last synced: 5 days ago
JSON representation
LLM Based ChatBot for Quantitative Carbon foorprint analysis
- Host: GitHub
- URL: https://github.com/bekcsys/LLM-Chatbot
- Owner: BeTKH
- Created: 2025-01-06T04:43:54.000Z (over 1 year ago)
- Default Branch: master
- Last Pushed: 2025-02-22T07:46:05.000Z (over 1 year ago)
- Last Synced: 2025-02-22T08:29:29.325Z (over 1 year ago)
- Topics: deepseek-r1, llama3, llms, mistral
- Language: Python
- Homepage:
- Size: 1.45 MB
- Stars: 0
- Watchers: 1
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
Awesome Lists containing this project
README
## ChatBot with LLMs
### Video Demo (click on the image)
[Watch video Demo](https://www.youtube.com/watch?v=EgKAD9C_Hks)
#### Output comparison from various models
### How to run the app
1. Setup virtual environment with [`pipenv`](https://pipenv.pypa.io/en/latest/installation.html)
2. Install dependencies
```
pipenv install
```
To install exact versions run:
```
pipenv install --ignore-pipfile
```
3. Download ollama (Two approaches:)
- form from [github](https://github.com/ollama/ollama?tab=readme-ov-file)
or
- from the [website](https://ollama.com/download/mac)
4. Download LLM models via ollama
Explore list of LLM models via ollama [here](https://ollama.com/library)
e.g. to download `llama3.2` run:
```
ollama run llama3.2
```
e.g. to download `deepseek-r1` run:
```
ollama run deepseek-r1
```
To download model with specific size e.g. 14B parameters = 9.0GB here is the [link](https://ollama.com/library/deepseek-r1:14b)
```
ollama run deepseek-r1:14b
```
5. Run the app
```
python app.py
```
OR
```
python3 app.py
```
### App challanges
- live rendering of outputs to front end (frontend problems)
- some models are rellay heavy, but accurate (smller models are less acurate)
- formatting of outputs e.g tables, codes, images
- performing quantitative analysis and visualizations & rendiering those to front end
### some goals
- add a drop down menu to select model
- format and clean generated quntitatve data to perform visualizations
- steering LLM output using feedback mechanism